Market Sizing Framework: Top-Down vs Bottom-Up, With Worked Examples
The market sizing framework is simple to state and hard to execute under pressure: pick top-down or bottom-up, announce your path, run clean numbers, then cross-check. McKinsey, BCG and Bain all test this skill, because it compresses everything they score into five minutes: structure, explicit assumptions, numerical fluency and business judgment. This guide explains when to use each approach, then solves one question both ways so you can see exactly how the two paths converge, followed by four more fully worked examples.
What Is a Market Sizing Question?
A market sizing or guesstimate question asks you to estimate the size of a market, a volume, or a count with limited data. Typical prompts: "How many coffee machines are sold in France each year?" or "How many tennis balls are sold in the US?"
Interviewers score you on structure, assumptions, math, and sanity-checking. A candidate within 50% of reality with a clean approach scores higher than one who guesses correctly without structure.
The Market Sizing Framework: Top-Down vs Bottom-Up
Every market sizing question can be tackled two ways. Top-down starts from a macro anchor, usually a population or a household count, and filters down with penetration rates and frequencies. Bottom-up starts from the supply side, units, outlets or transactions, and multiplies up. Strong candidates pick one, solve it cleanly, and cross-check with the other when time allows.
| Approach | Starts from | Best for |
|---|---|---|
| Top-down | Population or macro total | Consumer markets, frequency-based |
| Bottom-up | Units, outlets, or transactions | Infrastructure, supply-side questions |
The decision rule: if the question is about people doing or buying something, start from the people. If it is about physical capacity, outlets, machines or infrastructure, start from the units. When both paths are available, the one with fewer uncertain assumptions wins.
Insider Tip: State your approach explicitly. "I will take a top-down approach, starting from the number of French households." This single sentence earns structure points immediately.
Essential Anchor Numbers
Memorise a short list of anchors. You will reuse them in 80% of market sizing questions.
These are rounded working anchors, not live statistics. Interviewers expect memorable round numbers, not precision; state the vintage you are assuming.
| Anchor | France | Germany | UK | US |
|---|---|---|---|---|
| Population | 68M | 84M | 67M | 330M |
| Households | 30M | 41M | 28M | 130M |
| GDP | 2.8T EUR | 4.1T EUR | 3.1T GBP | 28T USD |
| Average salary | 38K EUR | 45K EUR | 35K GBP | 55K USD |
You should also know rough life expectancy at 80 years, average household size between two and three, and urbanisation rates in developed markets around 80%.
Worked Example 1: One Question, Solved Both Ways
Prompt: how many coffee machines are sold in France each year?
This is the rare question where both paths work well, which makes it the perfect illustration of the framework. Watch how two independent routes land on the same number.
The top-down path: start from households
- Anchor. France has about 30 million households.
- Penetration. Assume 75% own at least one coffee machine of some kind, filter, pod or espresso. That gives roughly 22 million machines in use.
- Replacement cycle. A machine lasts 5 to 7 years, call it 6. Annual replacement demand: 22 million divided by 6, about 3.7 million.
- Add first purchases. New households, second machines and gifts add roughly 10%, another 0.4 million.
- Total. About 4 million machines per year for the consumer market. Offices, hotels and restaurants would add a few hundred thousand units, so confirm the scope with your interviewer before including them.
The bottom-up path: start from the shelves
- Channels. Coffee machines are sold through grocery retail, electronics chains and online.
- Grocery. About 2,000 hypermarkets selling 8 machines a week and 6,000 supermarkets selling 2 a week: 28,000 units a week, roughly 1.4 million a year over 50 weeks.
- Electronics chains. About 1,500 stores at 15 machines a week: roughly 1.1 million a year.
- Online. Physical retail totals 2.5 million. If online takes about 40% of small appliance sales, physical is 60% of the market, so the total is 2.5 million divided by 0.6, about 4.2 million, with online at 1.7 million.
- Total. About 4 million machines per year.
Why the convergence matters
Top-down said 4.0 million. Bottom-up said 4.2 million. Two independent methods landing within 10% of each other is the strongest signal you can send in an interview: your estimate is defensible. If the two paths had diverged by a factor of three, one assumption would be broken, usually the penetration rate or the per-store volume, and finding which one is exactly the conversation the interviewer wants to have.
How McKinsey, BCG and Bain Test Market Sizing
Each firm wraps the same skill in a different format.
McKinsey runs interviewer-led cases and publishes official practice cases such as Beautify and Diconsa on its careers site, with the logic and suggested answers included. Sizing shows up as discrete quantitative questions inside those cases, so a fast, structured estimation reflex protects your time for the synthesis.
BCG runs candidate-led cases, described on its case interview preparation page. Market entry prompts, a BCG staple, usually begin with a sizing step: before debating share, you need the size of the pool. The quantitative sections of the BCG Casey online case reward the same reflexes.
Bain publishes practice cases on its case interview page, and market sizing is a frequent first-round exercise, either standalone or as the opening step of a larger case. Our Bain interview guide covers where it sits in the process.
Worked Example 2: Coffee Shops in Paris
Approach: bottom-up from demand.
Paris has around 2.2 million residents plus roughly 30 million annual tourists, which translates to about 2.5 million daily equivalents. Assume 50% of adults buy coffee from a shop on any given day, giving around 1.2 million daily coffee transactions.
A typical independent coffee shop sells around 200 coffees per day. Dividing 1.2 million daily transactions by 200 gives roughly 6,000 outlet-equivalents. Many of those transactions happen at chains and bakeries, so pure coffee shops are probably closer to half that, around 3,000.
Cross-check top-down: 2.2 million residents times 0.5 shops per 1,000 inhabitants gives 1,100, which is low because it ignores tourism. The bottom-up figure of around 3,000 is more defensible.
Worked Example 3: Tennis Balls Sold in the US Each Year
Approach: top-down by segment.
Start with three demand segments: recreational players, tournament and club play, and professional or training use.
- Recreational: 20 million occasional players times 6 balls per year equals 120 million
- Club and tournament: 5 million active players times 30 balls per year equals 150 million
- Professional and training: negligible at the total market level, maybe 10 million
Add cross-check demand from non-tennis uses such as dog toys and chair feet, which adds roughly 30 to 50 million. Total sits near 300 to 330 million balls per year.
This example shows the value of segmenting before multiplying. A single average would badly under or over-shoot.
Worked Example 4: New EV Charging Stations Needed in the UK by 2030
Approach: bottom-up from the car parc.
The UK has around 33 million passenger cars on the road. Assume 50% are electric by 2030, giving around 16 million EVs.
Industry rules of thumb suggest one public fast charger per 10 to 20 EVs, plus home chargers for around 60% of owners. The public charger need is 16 million divided by 15, or roughly 1 million public chargers. Working from a base of around 75,000 public chargers at the start of 2025, the gap is close to 900,000 new units by 2030.
Cross-check: that implies adding roughly 150,000 chargers per year, several times the recent pace of roughly 15,000 to 20,000 new chargers per year, so name that ramp-up assumption explicitly. Use these as working benchmarks rather than live figures; in a real interview, state the base year you are assuming.
Worked Example 5: Annual Revenue of the French Fitness Market
Approach: top-down from population.
France has 68 million people. Assume 10% are active gym members, giving 6.8 million members. Average membership is around 35 euros per month, or 420 euros per year. Core gym revenue is 6.8 million times 420, or roughly 2.9 billion euros.
Add adjacent revenue from boutique studios, personal training, and corporate wellness at roughly 40% of core. Total market is around 4 billion euros.
Cross-check bottom-up: France has roughly 4,500 gyms with average revenue of 800,000 euros each, giving 3.6 billion. Plus boutique and PT revenue, the convergence around 3.5 to 4 billion is robust.
Common Mistakes to Avoid
- Jumping into math without stating the approach
- Using a single average rather than segmenting
- Forgetting to sanity check with a different method
- Refusing to round or chasing false precision
- Ignoring the range and caveats at the end
Strong candidates finish with a range, such as "I estimate 3.5 to 4 billion euros, with the main sensitivity on membership rate and average price." This shows numerical humility and business judgment.
Pro Tip: End every sizing with one sentence on sensitivity. "If the membership rate is 8% rather than 10%, the market is closer to 3.2 billion." This almost always earns an extra point on structure.
Practice and Next Steps
Market sizing rewards reps. Aim for at least 20 varied questions across consumer, industrial, and infrastructure categories before your interview, and practice market sizing under interview pressure with an AI interviewer so the clock is part of every rep. Pair this with our case interview frameworks guide to tie sizing into broader cases, and with profitability cases, where sizing feeds the diagnosis, to see the estimate put to work.
FAQ
What is the market sizing framework?
It is the two-path structure behind every estimation. Top-down starts from a macro anchor such as population or households and filters down with rates. Bottom-up starts from the supply side, units, outlets or transactions, and multiplies up. Strong candidates announce the path, run it cleanly, then cross-check with the other.
Top-down or bottom-up: which is better?
Neither is universally better. Top-down works for consumer markets tied to population. Bottom-up works for supply-side or infrastructure questions. The best candidates pick one, solve it, and cross-check with the other.
How much time should a market sizing take?
Aim for three to five minutes. Anything longer suggests the structure is unclear.
Do I need to memorise country populations?
Yes for your target markets and major economies. These anchors appear in the majority of sizing prompts.
What if my number is off by a factor of two?
That is usually fine if your approach was clean and your assumptions were stated. Interviewers care about reasoning quality more than numerical precision.
Should I always cross-check with a second method?
If time permits, yes. Even a rough cross-check shows rigour and catches order-of-magnitude errors.
Further Reading
- McKinsey & Company, Interviewing: tips, practice cases and resources
- Bain & Company, Preparing for the Case Interview
- BCG, Consulting Case Study Interview Preparation
CasingLab offers adaptive case interviews with a McKinsey-style AI interviewer, 6-dimension scoring with proof points, and 500+ drills. CasingLab is now open. Start the free diagnostic and see your readiness score in five minutes.